seatrac_hackday_2024_code

Setup

filter out rare genes and normalize for CD8 and CD4 samples separately

For each dataset, I removed genes with expression in < 20% of samples and with < 10 CPM across samples

CD4 before-and-after removing rare genes

CD8 before-and-after removing rare genes

Voom normalization

CD4

CD8

Model logCFU vs gene

Just CD4_gran samples, association with logCFU

variable	fdr<0.05	fdr<0.1	fdr<0.2	fdr<0.3	fdr<0.4	fdr<0.5
(1 | subject)	744	936	1245	1492	1785	1992
logCFU	35	85	482	1658	2720	3997
total (nonredundant)	779	1015	1692	2988	4167	5430

Just CD4_gran samples, association with logCFU

variable	fdr<0.05	fdr<0.1	fdr<0.2	fdr<0.3	fdr<0.4	fdr<0.5
(1 | subject)	861	1202	1663	2106	2459	2850
logCFU	25	47	127	219	312	489
total (nonredundant)	886	1245	1780	2300	2730	3254

get top hits for low CFU-associated genes in CD4 and CD8

Just CD4_gran samples, genes with negative association with logCFU

$BFSP2

$LOC106998813

$LOC107001007

$LOC697141

$LOC698130

$LOC700872

$LOC702620

$LOC703283

$LOC706087

$LOC708140

$LOC709459

$LOC710859

$LOC712614

$LOC719289

$PSMD9

$SMIM12

$STOX1

Just CD8_gran samples, genes with negative association with logCFU

$ARHGAP19

$BTBD19

$C1H1orf198

$C7H14orf79

$C7H15orf62

$C9H10orf128

$C9H10orf90

$CFAP73

$ENKD1

$ENO3

$HEATR1

$KBTBD8

$LOC100429401

$LOC106997346

$LOC106998746

$LOC707027

$LOC720294

$PDCD1LG2

$RAD54L

$SFXN5

$SLC7A1

$STX3

look for agreement with Bromley

Sig negative corr in any cluster of Bromley, CD4_gran cells in Foreman (FDR < 0.2)

P-values represent interaction between group and logCFU on gene expression

$STOX1

Sig negative corr in any cluster of Bromley, CD8_gran cells in Foreman (FDR < 0.2)

P-values represent interaction between group and logCFU on gene expression

$STX3

$PDCD1LG2

$SLC7A1

$BTBD19

$HEATR1

$ENKD1

$C9H10orf128

$C1H1orf198

$SFXN5

$RAD54L

$ENO3

$C7H15orf62

$C7H14orf79

$ARHGAP19

$CFAP73